On 07/02/13 17:50, Jeroen Kransen wrote:
We are regularly doing a Dataset.replaceNamedModel(id, model), but it seems
like the underlying TDB files keep growing endlessly. I read this post
about someone having the same issue:
http://stackoverflow.com/questions/11088082/how-to-reduce-the-size-of-the-tdb-backed-jena-dataset
ReplaceNamedModel is their first case - delete then add.
Currently, index space is reused in the same JVM, but not across runs.
Node spaces is not reclaimed but nodes are reused.
OS issues matter - the meaning of "size" commands can be a bit weird on
some OS because the files start as sparse; if an area is touched they
become real space. How this is counted differs.
But I was wondering if there is a real solution, something like vacuuming
the TDB. If there is no solution, would SDB have the same issue?
SDB should clear up (use whatever the SQL DB provides) - it's unrelated
to TDB in that respect.
The current vacuum solution is to write up the TDB/DB and create a new one.
Jeroen
Andy
